The high component densities, clock rates and rise times that occur with Very High Speed Integrated Circuit (VHSIC) and Very Large Scale Integrated (VLSI) devices make the design concepts of the recent past inadequate. Some of these inadequacies stem from limitations inherent in the contemporary ceramic packaging materials that are used for substrates, thick film and the like. In order to fully appreciate the capabilities of the emerging technologies, more sophisticated electrical and mechanical requirements must be taken into consideration in the overall packaging, design decision. In other words, some requirements, particularly the electronic properties of the packaging material must be factored into the finalized VHSIC design.
Two primary packaging material properties which influence circuit performance are dielectric constant and dissipation factor. Of the two, dielectric constant is by far the more important since the dissipation factor's importance increases directly with the interconnect lengths. By reducing interconnect lengths a designer may deal more directly with the problem of the keeping the dielectric constant appropriate for the intended circuit applications.
The importance of having the proper dielectric constant is that the propagation delay, .sigma. is affected by the dielectric constant, .epsilon.' and can be expressed by the relationship that it is approximately equal to the square root of .epsilon.'. From this expression it is apparent that a significant reduction in the dielectric constant will have a useful impact on the speed at which the associated integrated circuits operate. For instance, to make a 33% decrease in signal delay, the dielectric constant must be reduced about 55% or, stated otherwise, the dielectric constant must be reduced from about 9 to 4.
A good many ceramic packaging devices including thick films and substrates are fabricated exclusively from alumina (Al.sub.2 O.sub.3) based systems. These ceramics often are manufactured by solid state sintering or glassy phase sintering and pure alumina substrates (up to 99.9% vol.) are manufactured without a glassy phase binder, although there is still this smallest percentage of tramp impurity glassy phase included. One of the advantages of pure alumina substrates is the virtual absence of a glassy, phase ensuring the highest possible thermal conductivity, which approaches approximately 35 W/mK. While alumina based systems have attractively high thermal conductivities and strengths, they possess an unattractively high dielectric constant equal to about 9. Since dielectric constant is a most important consideration in VHSIC designs, the relatively high dielectric constant negatively impacts logic speed and design in terms of impedance matching. Another limitation of the Al.sub.2 O.sub.3 based systems is that they must be fired at higher temperatures to promote the proper densification of the substrates. This is because in the absence of a glassy phase, firing temperatures approaching 1600.degree. C. are necessary. Even the more common 92% alumina (which actually has about 15 vol% of glass that is intentionally incorporated) has its firing and metallization temperature at about 1500.degree. C. The high firing temperatures, particularly in case of multilayer structures, require that the interconnecting conductors be metallized with refractory metals such as tungsten or molybdenum. As a consequence, higher line resistances must be tolerated. More desirable metals that have a higher conductivity, such as copper, silver or gold cannot be used since their continuity would be impaired during the application of the higher temperatures required for firing a substrate in which alumina is a major constituent.
Some of the contemporary developments use a dielectric filler, often Al.sub.2 O.sub.3 in a major glass fraction. Usually this glass fraction is over 50 vol% and may reach the 80 vol% figure.
A number of contemporary developments is based on the use of a glassy fraction composed of the borosilicate compositions, one of which is commercially available by the Corning Glass Works in its trademarked Pyrex series. These glasses have received acceptance because they are widely available commercially, time tested and low in dielectric constant.
Most of the borosilicate glasses have only a small fraction of aluminum oxide and do not contain calcium oxide as a major phase in the glassy binder. A high volume fraction of the glassy binder has been used in a number of the contemporary compositions because the glass has a relatively high viscosity which retards thermal densification unless incorporated as the major phase. It appears that greater than 50 vol% glass is necessary with Pyrex type borosilicates to bring the firing temperature down below 1000.degree. C. However, such a large glass fraction reduces the strength of the final material. As a consequence, it is desirable to reduce the large glass fraction. Such a reduction, however, is impossible when a low firing temperature is needed with the borosilicate glasses.
A system of glasses CaO--BaO--Al.sub.2 O.sub.3 was discussed by A. E. Owen in his paper entitled "Properties of Glasses in the System" CaO--BaO--Al.sub.2 O.sub.3 appearing in Physic and Chemistry of Glasses, volume 2, #3, June, 1961. The glasses Owen discussed are quite different in composition and most different in viscosity as compared to the Pyrex type borosilicate. The lower viscosity of the CaO--BaO--Al.sub.2 O.sub.3 glasses provides a property which greatly aids sintering and this property enables the use of as little as 25 vol% in the glassy phase in this invention to enable the sintering below 1000.degree. C. SiO.sub.2 can be added to the overall composition to improve its moisture stability and was added as a minor fraction in one batch of this invention whereas the Pyrex glasses employ SiO.sub.2 as a major component. A disadvantage of including SiO.sub.2 as a major portion is that evidence suggests that nucleation, or devitrification of high expansion cristobalite is likely to occur. The sudden volume change of cristobalite around 200.degree. C. would destroy circuit elements bonded to the substrate.
Other attempts have been made to develop low firing, high conductivity metallized, user-fired tape systems for fabrication of single and multilayer substrates. These systems are mostly glass based systems with dielectric fillers and have dielectric constants of approximately 8. This high dielectric constant is not suitable for the present day VHSIC and VLSI electronic circuit applications. One other system worthy of mention is a true glass ceramic system in which a glass is crystallized after densification. This system has the proportion of the glassy phase in excess of 50% in order to keep the firing temperature low. However as with the other systems, this high a percentage of the glassy phase compromises strength and reliability. The trade off of compromising strength and reliability of the substrate to accommodate lower temperature firing is made routinely by designers. This allows higher conductivity metallizations.
There is a strong need for a glassy binder which could serve to consolidate ceramic fillers at relatively low temperatures and low volume fractions. Such a glassy binder could serve not only as a glassy phase in a substrate package and multilayer applications, but in thick film applications as well where high conductivity metallization are desirable.
A continuing need exists in the state-of-the-art for a low dielectric constant of about 4, co-fired multilayered alumina compatible substrate material having a minor glass phase and which will fire below 1000.degree. C. to allow high conductivity metallization and which does not overly compromise its inherent strength.